JPS6281616A - Focus position detecting device - Google Patents

Abstract

PURPOSE:To perform highly stable focus detection with high precision by photo detecting only reflected light from a body to be measured which is generated with two kinds of striped pattern light having a periodically varying light and shade part with arrayed photosensors through a cylindrical lens. CONSTITUTION:Illumination light becomes two kinds of stripe pattern light having the periodically varying light and shade part through striped pattern 3 for a front and a rear focus to illuminate the object body through a half-mirror 4a which is so sized and positioned as to exert any influence upon light from other optical systems. Then, the reflected light from the object body is transmitted through the mirror 4a and then the reflected light having a high SN ratio has the stripe pattern compressed lengthwise through a cylindrical lens 11. This compressed pattern light is photodetected by the arrayed sensors 7 with a good SN ratio without projection and only light and shade information on picture elements at the center part of the striped pattern is extracted so that the enlargement image width of either light or dark striped pattern coincides with the integral multiple width of the picture elements of the sensors 7. Therefore,the light and shade information is detected excellently without being affected by a difference in reflection factor, etc., due to the irredular state of the surface to be measured and the material, so that focus detection is performed highly stably with high precision.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention is a voice point position detection device suitable for an automatic focusing device that automatically focuses on a patterned LSI wafer surface.

[Background of the invention]

An example of a focal position detection device using a periodic brightness pattern projection method is disclosed in Japanese Patent Laid-Open No. 70540/1983. The former method will be explained using FIGS. 19 to 27.

Figure 19 shows its basic configuration, and the striped pattern is divided into two on a plane along the optical axis so that the image plane is slightly shifted in front and behind the focal point #LVc of the sample surface observation optical system. It is arranged as follows. The illumination light emitted from the light source 1 passes through the lens 2, reaches the striped pattern section 3, is reflected by the semi-transparent mirror 4, enters the imaging lens 5, and forms a front focal point on the surface of the sample 60, which will be described later. Striped pattern for use and back focus S&ISb
to project. As shown in FIG. 2, the striped pattern is composed of a repeating pattern of light blocking portions 9 and light passing portions 10. The light shielding part 9 and the light passing part 10- set constitute one detection brightness and darkness pattern.

As shown in the figure, the striped pattern is divided into two parts 3a and 3b along a plane perpendicular to the center of the optical axis. Front focus stripe pattern 3
a, the focal point positions of the back focal striped pattern 3b are provided slightly toward the front and rear of the focal point position of the array sensor 7.

However, the pre- and post-focus striped patterns are arranged so that the amounts of deviation are equal. Next, the reflected light corresponding to the optical properties of the surface of the sample to be measured is focused by the imaging lens 5 and passes through the semi-transmissive mirror 4, after which a striped pattern is projected onto the play-shaped sensor 7. As shown in FIG. 3, the array sensor has photoelectric elements 71 arranged in an IJ sock shape. The correspondence relationship between the photoelectric elements and the imaging stripe pattern on the arrayed sensor surface is as shown in the figure.
One bright (dark) pattern corresponds to a column.

Next, the output signal from the array sensor 7 is shown in FIG. As shown in the figure, an output signal matching the repetition period of the chiaroscuro projection pattern is obtained.

The output signal is subjected to the following processing in the signal processing circuit 8 to detect the focal point position. The difference between the bright output voltage Vl and the dark output voltage V is divided by the bright output voltage V.

This division result is defined as contrast C. Next, the sum of contrasts ΣC1, obtained from the brightness information of the front focus striped pattern, and the sum of contrasts ΣC8, obtained from the brightness information of the back focus striped pattern, are as shown in FIG. 25, and ΣCF and ΣCo The position where the two become equal is the in-focus position.

This device has the following problems. (1) The amount of light obtained as brightness information of the striped pattern is limited by the size of the array sensor, resulting in insufficient light amount, resulting in insufficient S/N ratio and unstable operation. for example,
An imaging lens 5, a 50x objective lens, and an array sensor 7
The size of the light receiving surface is 4 m (L) x 2 wm (
W) (see Fig. 21), the size of the entire striped pattern is uniquely determined as 4 m (/,) x 2 m (-).

(2) Since one striped pattern corresponds to one row of the arrayed sensor in the W direction, the operation becomes unstable due to mutual positional deviation between the striped pattern and the arrayed sensor in the LCL direction. In addition, when the object to be measured is defocused, the difference between the bright output voltage v1 and the dark output voltage v2 rapidly approaches zero, so it is necessary to have a wide operating range m (Fig. 23) for detecting the amount of defocus. I can't.

(3) First, we will discuss the fact that the brightness of reflected light varies depending on the surface condition of the object to be measured.

1) Variations due to surface irregularities For example, in an LSI wafer, as shown in FIG. 24, there are areas where the pattern is dense and areas where it is sparse. In the former, the illumination light is scattered at the edge of the pattern, making it dark υ, while in the latter, it is not scattered and is therefore bright.

11) Variations due to differences in reflectance of surface materials Similarly to 1), an LSI wafer will be taken as an example. As shown in FIG. 7, in an LSI wafer, there are many layers on a Si substrate, and each material has a different reflectance. Generally, they are classified as follows.

3i, A1... Si3N4 r Po1y-8t... Low reflectance. For transparent layers such as SiO2, the surface brightness depends on the reflectance of the underlying layer, but even if the underlying layer is the same, the brightness will be 5if.
The surface brightness varies depending on the thickness of t.

Therefore, if a part of the projected stripe pattern overlaps the boundary where brightness variations occur as shown in FIGS. 26(a) and 26(b), the detected brightness signal will also vary accordingly. However, with this device, the detected brightness and darkness signals at the focused point at the measurement location shown in Figure (a) are as follows.
The result is the same as the detection signal at the time of defocusing at a location with uniform surface brightness (FIG. 10(C)), resulting in false detection.

For example, the positional relationship between the sample image detection field and the projected striped pattern is as shown in FIG.
is inserted into a part of the optical path (see FIG. 18), and the optical path of the valence pattern and the optical path of the sample image detection section cannot be separated.

(4) Since the semi-transmissive mirror is inserted throughout the cross-section of the optical path, the amount of light for observing the sample surface is reduced by the reflectance of the semi-transmissive mirror.

Therefore, when the sample surface observation light is photoelectrically converted, the S/N of the sample surface observation signal decreases.

[Purpose of the invention]

SUMMARY OF THE INVENTION An object of the present invention is to prevent the occurrence of focal position detection errors due to the above-mentioned problems and to provide a stable and highly accurate focal position detection device that is not affected by the condition of the sample surface.

[Summary of the invention]

The present invention improves the conventional method of detecting the focus position without correcting variations in brightness on the sample surface as follows, and enables highly accurate and highly stable focus position detection.

(1) By installing a cylindrical lens in front of the sensor and performing optical image compression in the longitudinal direction of the striped pattern, brightness information can be stably detected with a high S/N ratio.

(2) As shown in Figure 16, multiple sensors (three pixels in this case) are associated with one projection stripe pattern, and only the sensor output signal in the center is sampled to stabilize the detection signal. 〇(3) Multiple projection stripe patterns as the 17th
As shown in the figure, a set of striped patterns for front focus and back focus (
or the first set) are arranged alternately, and furthermore, the projected fringe pattern is made to have an angle of 45° with respect to the pattern on the surface of the object to be measured, and focusing is performed according to the above-mentioned variation in brightness on the surface of the object to be measured; to avoid detection errors.

(4) A part of the array sensor is used to measure the specularly reflected light of the striped pattern projection part on the surface of the object to be measured, and the brightness signal is corrected using the constant value of track 1, and the variation in brightness on the surface of the object to be measured is corrected. This prevents focus position detection errors from occurring.

(5) I! As shown in Figure 18, the striped pattern projection part and the sample image ejection part? Keep the semi-transparent mirror for striped pattern projection at a sufficient distance.
It was made into a large version that only projected three patterns. Therefore, try
Since the amount of light for detecting the sample image reaches the detector without being reduced by the semi-transmissive mirror 4a, the S/N of the sample image detection signal is improved.

In addition, since no extra stray light enters the array sensor for striped pattern brightness information detection, the S/N of the brightness information detection signal is
This also improves performance and stabilizes operation.

[Implementation sequence of the invention]

An example in which the present invention is applied to a wafer foreign matter inspection device will be described. The overall configuration is shown in Figure 1.

This device consists of an on-wafer foreign object detection section and a focus position detection section. The wafer is fixed on two stages 27 and scanned in the X-Y directions as shown in FIG. During foreign object detection, a focus shift occurs due to the waviness of the wafer 6, but this has a negative effect on the foreign object detection performance, so it is necessary to always maintain the focal position.

Therefore, if the focus position detection method described above is used, as shown in the figure, the focus position is detected in parallel with the foreign object inspection and feedback is applied to two stages with a two-direction drive mechanism, and the focus shift can be detected in real time during the foreign object inspection. Can be corrected. First, an overview of the device will be explained. The foreign object detection section includes an inter-image system 29(a)(b), a detection system 22, a signal processing system 25, a foreign object memory 24, and a display 25. Further, the wafer 6 is vacuum-adsorbed on the two stages 27 and scanned in the X-Y direction by the X-Y stage control circuit 26 and the X-Y stage 28. During foreign object inspection, a focus shift occurs due to the waviness of the wafer 6, but this affects the foreign object detection performance, so it is essential to maintain the focal position at all times. Therefore, the focus position detection method of the present invention is applied to the focus position detection units 7, 7.
00, 8, the focus position is detected in parallel with the foreign object inspection and feedback is applied from the 2 driver 5o to the Z stage 27, as shown in the figure, so that the focus shift can be corrected in real time.

Next, the foreign object detection section will be explained. In this detection method, as shown in Figure 3, a He-Ne laser 2 is emitted from diagonally above the wafer.
9a, 2? b is illuminated from both sides in the
Pin Si photodiode linear sensor 22
After photoelectrically converting the foreign object signal and storing the foreign object signal in the foreign object memo IJ241C and scanning the entire wafer, the foreign object position is displayed on the display 25. As shown in the figure, since the foreign matter inspection section and the striped pattern projection section are separated by a distance a on the wafer, their respective optical paths can be easily separated using the semi-transmissive mirror 4a.

The detection signal from the foreign object is several tens of mV, as shown in Figure 4.
In the conventional method, the optical path of the foreign object detection system and the focal position detection system were not separated, so some foreign objects of about 1 μm could be detected with a sufficient S/N ratio with the noise. Occasionally, foreign objects were overlooked. However, in the present invention, since the two optical paths are separated as described above, the S/N of the foreign object signal is twice as large as that of the conventional method (semi-transmissive mirror 4
(when the transmission is 50% and the reflection is 50%), and all foreign particles of about 1 μm can be stably detected.

This separation is performed using an illumination method specific to the foreign object inspection device (i.e., instead of inspecting with the illumination light 1, the external illumination light 29a, 29b
Applicable only to lighting

Next, the autofocus system will be explained.

FIG. 1 shows the basic configuration of the present invention. The illumination light emitted from the light source 1 has a striped pattern 3a.

3b, and projects the striped pattern onto the surface of the object to be measured 60. The striped pattern is as shown in FIG.
The light passing section 1o and the light shielding section 9 constitute one set of front (rear) focal position detection patterns. The reflected light reflected from the sample surface is condensed by the imaging lens 5, further compressed only in the longitudinal direction of the striped pattern by the cylindrical lens 11, and the striped pattern is projected onto the array sensor 7. The detection signal detected by the array sensor is as shown in FIG. The relationship between out-of-focus and detection signals is also shown in the figure.
'Next, the effects of the features of this patent will be explained.

(1) Effect of cylindrical lens In this embodiment, as shown in FIG. 6, a cylindrical lens was installed so as to compress the length uP) of the enlarged projected fringe pattern at the sensor position to the sensor length W). As a result, the effective amount of light detected by the array sensor increases by 1/W times compared to when no cylindrical lens is installed, and the S/N of the detection signal increases.
The ratio can be improved.

(2) Effect of making multiple picture elements correspond to one striped pattern In this embodiment, as shown in FIG. 16, three pixels of the sensor are made to correspond to one striped pattern, and only the signal in the center is detected. . FIG. 7 shows the light intensity distribution when a striped pattern is imaged at the sensor position. FIG. 7(sL) shows the light intensity distribution at the sample focus position, and FIG. 7(b) shows the light intensity distribution at the time of defocus. In the conventional example, 1 stripe pattern is used for each stripe pattern.
Since the picture elements were made to correspond, St and Sx in the shaded area shown in FIG. 7(a) are all photoelectrically converted and the signal output V1. V
, becomes. In this case, the contrast calculation result (C above)
quickly approaches zero with respect to the amount of positional deviation, as shown in FIG. 8(a).

On the other hand, when three pixels are associated, although the amount of light photoelectrically converted by the central one pixel decreases, S't in Fig. 7 (It),
Since S≦ is photoelectrically converted, the contrast calculation result is the 20th
As shown in the figure, it slowly approaches zero. Therefore, the operation is more stable when one stripe pattern is made to correspond to a plurality of sensor picture elements and only No. 1 in the center is detected than when one stripe pattern is made to correspond to one sensor pixel. In addition, in this method, the relative position of the sensor and the striped pattern in the L (t) direction has more margin than in the conventional method, and the position of the striped pattern can be easily adjusted.

(5) The effect of alternately arranging the front focus stripe pattern and the back focus stripe pattern, and the effect of tilting the stripe pattern by 45 degrees with respect to the pattern of the object to be measured.

As a result of alternately arranging the front focus stripe patterns and the back focus stripe patterns, the area onto which each set of focus stripe patterns is projected is subdivided by the number of repetitions. Therefore, even if a portion of the projected fringe pattern overlaps the boundary portion where brightness variations occur on the sample surface as described above, the influence of the brightness variations on the detection signal is reduced. Further, as shown in FIG. 9, in the conventional method, erroneous detection occurs when the repetition period of the wafer pattern matches the repetition period of the projected stripe pattern. In the present invention, we have focused on the fact that most of the patterns on LSI wafers are orthogonal patterns, and by tilting the projected stripe pattern by 45'' with respect to these patterns, it has become possible to prevent the above-mentioned erroneous detection.

(4) Effect of brightness correction First, we will discuss the division of the sensor. In this embodiment, both ends of the arrayed sensor shown in FIG. 10 are used to measure variations in brightness on the sample surface, and the remaining parts are used to detect brightness information of the projected fringe pattern. Various methods of dividing the sensor can be considered, but any method that simultaneously detects the brightness information of the projected stripe pattern and the brightness variation of the sample surface with the same sensor is within the scope of the present patent claim. The output signal from the sensor is as shown in Figure 11. FIG. 5A shows a striped pattern brightness information detection signal, and the bright level and dark level of the signal vary depending on the variation in brightness of the sample surface. Next, the results of measuring the variations in brightness are shown in FIG. 2(b). Here, if the surface brightness variation measurement signal Q)) is used to divide the striped pattern brightness information measurement signal (a), the same figure (
As shown in G), only the brightness information (normalized) of the projected fringe pattern, which removes the influence of variations in brightness on the sample surface, can be detected.

(5) Effect of separating the optical path into the sample surface observation section and the striped pattern projection section An example of application to a wafer foreign matter detection device to be described later will be described. In this device, the center of the field of view is used to detect scattered light from foreign objects on the sample, as shown in FIG. 3, and a striped pattern is projected on the periphery.

Figure 12 (a) shows a foreign object detection signal when a semi-transmissive mirror is inserted into the entire optical path cross-section of the detection system, with 50 reflections and 5 exposures inserted, and Fig. 12 (b) shows a foreign object detection signal when the same semi-transmissive mirror is inserted only in the striped pattern optical path. Shows the foreign object detection signal when inserted. As can be seen from the figure, the foreign object detection signal increases by the reflectance of the semi-transmissive mirror, and even foreign objects whose S/N ratio of foreign object signal to electrical noise signal was near the detection limit in the conventional method can be detected by S increases by nearly twice, allowing stable detection of foreign objects.

Next, the signal processing system will be explained. The entire circuit configuration is shown in FIG. 25. First, the outputs of the array sensor are taken out serially, and the brightness and darkness signals and brightness variation measurement signals are separately added by adders 13 and 15. The level of the brightness measurement signal is corrected by the amplifier circuit 14.

The brightness signal is then divided by the brightness signal and normalized. Next, the normalized bright/dark signal is sampled by a sample-and-hold circuit, and is divided into a bright part output signal and a dark part output signal. The divided signals are processed by a contrast calculation circuit (difference) 19.
Outputs the difference between bright and dark output as contrast,
In the next differential output calculation circuit, the sum ΣCF of the front focus contrast and the sum ΣCB of the rear focus contrast are determined by adders 205 and 206, respectively, the difference between ΣC2 and ΣC3 is determined by the difference circuit 207, and the results are Output as differential output. FIG. 26 shows a timing chart of the Crozu pulse used in this signal processing system. First, the array sensor is scanned by Cpo and outputs a signal serially.

In this embodiment, 3 pixels of the sensor correspond to one striped pattern, and a total of 8 sets of striped patterns are projected, so 48 pulses are required, and one scan is performed with a total of 49 pulses including reset pulse 1. Let's do it. The light and dark signals are extracted at Cpt*Cp2, and the timing is set so that only the light and dark information at the center 4 of the striped pattern is extracted. Cps, C2
0 (Cps, Cps) is a timing pulse for latching the front (back) focus contrast.

FIG. 15 shows the results of focal position detection using this method. The broken line in the figure is the detection result by the conventional method. In this way, it is possible to detect the focal position with high precision and a wide operating range.

As described above, an on-wafer foreign object detection device with an automatic focusing mechanism that operates stably has been realized.

〔Effect of the invention〕

As explained above, according to the present invention, it is possible to detect the focal position with high accuracy even on a sample with uneven brightness on the sample surface. Furthermore, since the light intensity of the foreign object detection system is not reduced, the S/N ratio of detection signals of other observation systems can be improved. Therefore, when applied to a wafer foreign matter detection device, the foreign matter detection performance is also improved.

[Brief explanation of drawings]

Fig. 1 is a basic configuration diagram of an embodiment of the present invention, Fig. 2 is a diagram explaining the wafer scanning method of the embodiment, Fig. 3 is a diagram showing the configuration of the optical system of the embodiment, and Fig. 4 is a diagram showing foreign matter. FIG. 5 is a diagram showing a graph of the detection signal,
Fig. 6 is an optical path diagram when a cylindrical lens is inserted, Fig. 7 is a graph explaining the light intensity distribution, the 8th front view, Fig. 11 is a graph showing the focal position detection signal, 1''2
Figure - is a graph showing the detection signal from a foreign object, Figure 13 is a configuration diagram of the signal processing system, Figure 1+ is a clock timing chart, Figure 15 is a graph showing the relationship between differential output and focus shift, Figure 16 is a diagram showing the positional relationship between the stripe pattern of the detection method of the present invention and the array sensor, Figure 19 is a diagram of the stripe pattern of the detection method and the configuration of the illumination section, and Figure 1g explains the optical path of the detection method of the present invention. Figure 19 is a diagram of the optical system configuration of the conventional detection method.
Fig. 20 is a diagram showing the stripe pattern arrangement of the conventional detection method, Fig. 21 is a diagram showing the relationship between the stripe pattern of the conventional detection method and the sensor position, and Fig. 22 is a diagram showing the relationship between the brightness detection signal and defocus of the conventional detection method. 23 is a graph showing the relationship between detected contrast and defocus, and a graph explaining the operating range m. The second figure is a surface view of the LSI wafer, and Figure 25 is the LS
A cross-sectional view of the I wafer, FIG. 26(a) is a graph showing the detection signal at the time of defocusing of the measurement point that causes false detection, and FIG. 27 shows the positional relationship between the stripe pattern projection part and the observation part within the field of view. It is a diagram. 1... Light source for illumination, 2.29... Condenser lens, 3... Striped pattern portion, 3a... Striped pattern for front focus, 3b... Striped pattern for rear focus, 4.4a...・Semi-transparent mirror, 5...Objective lens,
6... Sample, 7... Array sensor,
71...Sensor picture element, 8...Signal processing circuit,
9...One striped pattern, 10...Bright part of the striped pattern,
11... Cylindrical lens, 13, 15... Adder, 14... Amplifier, 1
6...Divider, 17.18...Sample & hold circuit, 19...
Differential device, 20... Differential output calculation circuit, 201
~240...Latch circuit, 205.206...Adder, 207...Differentiator,
21... Total reflection mirror (or semi-transmission mirror), 22... Foreign object detector, 23... Foreign object signal processing circuit, 24...
Foreign object memory, 25...Display unit, 26...X-Y
Stage drive circuit, 27--Z stage, 28. XY stage, 29
a l b...Lighting for foreign object detection, 50...Z stage driver, 700...Brightness information detection unit, 291...Condenser lens. ,·debt

Claims (1)

[Claims] 1. A first imaging optical system that projects and images a striped pattern of periodically repeating bright and dark areas onto an object to be measured; a second imaging optical system that forms an image on an array sensor made of elements; and a second imaging optical system for detecting the focal position of the object to be measured using brightness information of the striped pattern obtained by photoelectric conversion by the array sensor. A focal position detection device characterized by comprising an electric circuit. 2. The focal position detection device according to claim 1, wherein a cylindrical lens is installed in the second imaging optical system so as to compress each of the striped patterns in the longitudinal direction. 3. Match the enlarged image width of the striped pattern in either the bright area or the dark area with the integral multiple width of the picture elements of the array sensor, and extract only the brightness information of the picture elements in the center of each striped pattern. A focal position detection device according to claim 1, characterized in that: 4. A plurality of sets of the stripe patterns are formed, with respect to a predetermined focal point position of the first optical system, a stripe pattern having an imaging position in front (front focus stripe pattern) and an imaging position having an imaging position in the rear. Claim 1, characterized in that the image is divided into stripe patterns (rear focus stripe patterns), and the front focus stripe patterns and the back focus stripe patterns are arranged so as to alternately repeat the image formation at the imaging position. The focal position detection device according to item 1. 5. The array sensor is used to simultaneously extract brightness information of the striped pattern and measure the reflectance of the surface of the object to be measured, and the former detection signal is corrected based on the signal measurement result, and the reflectance of the surface of the object to be measured is corrected. 2. The focus position detection device according to claim 1, wherein the influence of variations in the focus position is removed. 6. Project the projected fringe pattern around the observation field of the object to be measured, use the center of the optical axis to detect the observation light of the object to be measured, and insert a semi-transparent mirror into a part of the observation field of view to see the projected fringe pattern. 2. The focal position detection device according to claim 1, wherein the optical path and the optical path of the object observation system to be measured are separated to increase the amount of light of both.